Spatial Modeling of Deep-Sea Ferromanganese Nodules With Limited Data Using Neural Networks

Vishnu, Hari; Kalyan, Bharath; Ganesan, Varadarajan

doi:10.1109/joe.2017.2752757

Cited by 14 publications

(5 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Regarding the bathymetric maps, the accurate and detailed reconstruction of the seafloor bathymetry at meterscale resolution enables to use bathymetry and its derivatives as source data layers within a high-resolution RF model. These data should have high-quality characteristics, as the presence of acquisition artefacts may affect the robustness of the modeling procedure (Preston, 2009;Herkül et al, 2017). The combined use of cameras as the DeepSurveyCamera (Kwasnitschka et al, 2016) for acquiring high-resolution photographs and an automated analysis with a state-of-theart algorithm (Schoening et al, 2017a) provide essential quantitative information about the distribution of Mn nodules.…”

Section: Discussionmentioning

confidence: 99%

Quantitative mapping and predictive modeling of Mn nodules' distribution from hydroacoustic and optical AUV data linked by random forests machine learning

et al. 2018

View full text Add to dashboard Cite

Abstract. In this study, high-resolution bathymetric multibeam and optical image data, both obtained within the Belgian manganese (Mn) nodule mining license area by the autonomous underwater vehicle (AUV) Abyss, were combined in order to create a predictive random forests (RF) machine learning model. AUV bathymetry reveals small-scale terrain variations, allowing slope estimations and calculation of bathymetric derivatives such as slope, curvature, and ruggedness. Optical AUV imagery provides quantitative information regarding the distribution (number and median size) of Mn nodules. Within the area considered in this study, Mn nodules show a heterogeneous and spatially clustered pattern, and their number per square meter is negatively correlated with their median size. A prediction of the number of Mn nodules was achieved by combining information derived from the acoustic and optical data using a RF model. This model was tuned by examining the influence of the training set size, the number of growing trees (ntree), and the number of predictor variables to be randomly selected at each node (mtry) on the RF prediction accuracy. The use of larger training data sets with higher ntree and mtry values increases the accuracy. To estimate the Mn-nodule abundance, these predictions were linked to ground-truth data acquired by box coring. Linking optical and hydroacoustic data revealed a nonlinear relationship between the Mn-nodule distribution and topographic characteristics. This highlights the importance of a detailed terrain reconstruction for a predictive modeling of Mn-nodule abundance. In addition, this study underlines the necessity of a sufficient spatial distribution of the optical data to provide reliable modeling input for the RF.

show abstract

Section: Discussionmentioning

confidence: 99%

Quantitative mapping and predictive modeling of Mn nodules' distribution from hydroacoustic and optical AUV data linked by random forests machine learning

et al. 2018

View full text Add to dashboard Cite

show abstract

“…Data gathered by sample collection and highresolution acoustic surveys are then fed into multivariate (geo)statistical models to predict nodule abundance and metal resources over large areas 5,78,79,83,84 . Such predictive models are based on artificial neural networks or random forests, as well as on classical geostatistics, such as variography and kriging, in order to assess the quantity of the resource 5,83,[85][86][87] . Once the economic potential of a nodule field has been determined, then, if favourable, the operations will move on to mining.…”

Section: Unique (Or Favourable) Characteristics Of Deep-ocean Miningmentioning

confidence: 99%

Deep-ocean polymetallic nodules as a resource for critical materials

Hein

Koschinsky

Kühn

2020

Nat Rev Earth Environ

225

127

View full text Add to dashboard Cite

Deep-ocean polymetallic nodules (also known as manganese nodules) are composed of iron and manganese oxides that accrete around a nucleus on the vast abyssal plains of the global ocean 1-6. Polymetallic nodules vary in diameter from less than one to tens of centi metres and acquire economically interesting quantities of critical metals (metals that are essential to the security and economic wellbeing of a nation) from ocean water and/or sediment pore waters. As a result, the enormous quantities of these nodules-which are conservatively estimated to total 21 billion dry tons in the Clarion-Clipperton Zone 7,8 (CCZ; located in the northeast equatorial Pacific Ocean) alone (Fig. 1)-host considerable tonnages of critical metals that are essential for green energy, vehicles and infrastructure, as well as for new technology and military applications. This reservoir of critical metals has, therefore, triggered interest in new deep-ocean mining operations. However, the potential environmental impacts of such activities are of great concern and have hastened extensive research and evaluation in this area 9-11. Polymetallic nodules were first discovered on the seabed at a depth of approximately 4,300-5,500 m during the 1872-1876 voyage of the HMS Challenger to study the deep ocean 12. A number of nodules, which contained numerous shark teeth and cetacean ear bones, were collected from the western end of the CCZ nodule field 12. Early hypotheses suggested that polymetallic nodules form by alteration of volcanic rocks and grains, especially glass, which are transformed into manganese carbonates and, finally, into oxides that are deposited from solution to form nodules on water-rich sediments 12-14. However, later work-aided in part by the 1957-1958 International Geophysical Year-revealed the general source of the metals (seawater and sediment pore water) and the sorption processes that might be involved in the acquisition of minor metals by nodules 15-20. Economic interest in the deep-ocean polymetallic nodules developed in the 1960s 15,16 , which subsequently led to the formation of consortia in Germany,

show abstract

“…The spatial/cluster-blocking integration within the model training was performed with the CAST R package [130]. (10) is the optimum number in CLARA clustering, which is relatively high for the small area (≈17 km 2 ), indicating considerable spatial heterogeneity. The central and northern parts have higher variability.…”

Section: Feature Space K-fold Clustering Cross-validationmentioning

confidence: 99%

“…Multibeam echosounder (MBES) data (bathymetry and backscatter), seafloor lithology, environmental information (e.g., organic carbon content), and ground-truth information (photos, information from box-corer sampling) have been analyzed with machine learning Minerals 2021, 11, 1172 2 of 33 (ML) methods, providing the spatial distribution of PMN [7,[10][11][12][13][14][15]. However, the influence of spatial autocorrelation (SAC) on ML modeling has not been considered.…”

Section: Introductionmentioning

confidence: 99%

Importance of Spatial Autocorrelation in Machine Learning Modeling of Polymetallic Nodules, Model Uncertainty and Transferability at Local Scale

Gazis

Greinert

2021

Minerals

View full text Add to dashboard Cite

Machine learning spatial modeling is used for mapping the distribution of deep-sea polymetallic nodules (PMN). However, the presence and influence of spatial autocorrelation (SAC) have not been extensively studied. SAC can provide information regarding the variable selection before modeling, and it results in erroneous validation performance when ignored. ML models are also problematic when applied in areas far away from the initial training locations, especially if the (new) area to be predicted covers another feature space. Here, we study the spatial distribution of PMN in a geomorphologically heterogeneous area of the Peru Basin, where SAC of PMN exists. The local Moran’s I analysis showed that there are areas with a significantly higher or lower number of PMN, associated with different backscatter values, aspect orientation, and seafloor geomorphological characteristics. A quantile regression forests (QRF) model is used using three cross-validation (CV) techniques (random-, spatial-, and cluster-blocking). We used the recently proposed “Area of Applicability” method to quantify the geographical areas where feature space extrapolation occurs. The results show that QRF predicts well in morphologically similar areas, with spatial block cross-validation being the least unbiased method. Conversely, random-CV overestimates the prediction performance. Under new conditions, the model transferability is reduced even on local scales, highlighting the need for spatial model-based dissimilarity analysis and transferability assessment in new areas.

show abstract

Spatial Modeling of Deep-Sea Ferromanganese Nodules With Limited Data Using Neural Networks

Cited by 14 publications

References 39 publications

Quantitative mapping and predictive modeling of Mn nodules' distribution from hydroacoustic and optical AUV data linked by random forests machine learning

Quantitative mapping and predictive modeling of Mn nodules' distribution from hydroacoustic and optical AUV data linked by random forests machine learning

Deep-ocean polymetallic nodules as a resource for critical materials

Importance of Spatial Autocorrelation in Machine Learning Modeling of Polymetallic Nodules, Model Uncertainty and Transferability at Local Scale

Contact Info

Product

Resources

About